cc #2223
BREAKING CHANGE: The HTTP server code is now an optional feature. To
enable the server, add `features = ["server"]` to the dependency in
your `Cargo.toml`.
cc #2223
BREAKING CHANGE: The HTTP client of hyper is now an optional feature. To
enable the client, add `features = ["client"]` to the dependency in
your `Cargo.toml`.
cc #2251
BREAKING CHANGE: This puts all HTTP/1 methods and support behind an
`http1` cargo feature, which will not be enabled by default. To use
HTTP/1, add `features = ["http1"]` to the hyper dependency in your
`Cargo.toml`.
The current implementation of `drain` uses a `tokio::sync::watch`
channel to send the shutdown signal, and a `tokio::sync::mpsc` to signal
when all draining tasks have completed. No data is ever actually sent on
the MPSC; instead, it is simply used to notify the task that signalled
the drain when all draining tasks have been dropped.
Tokio 0.3's `watch::Sender` has a `closed` method that can be used to
await the dropping of all receivers. This can be used instead of the
MPSC channel. This commit updates `drain` to use `watch::Sender::closed`
instead. This has fewer moving parts, and may have slightly less
overhead (as it doesn't require additional allocation forthe MPSC which
is never actually used).
Signed-off-by: Eliza Weisman <eliza@buoyant.io>
cc #2251
BREAKING CHANGE: This puts all HTTP/2 methods and support behind an
`http2` cargo feature, which will not be enabled by default. To use
HTTP/2, add `features = ["http2"]` to the hyper dependency in your
`Cargo.toml`.
`KeepAliveState` did not transition from `PingSent` to `Scheduled` after
a pong was received. This prevented more than one ping to be sent by the
server. This fix checks if `ping_sent_at` has already been cleared by
`Ponger::poll` when `KeepAliveState::PingSent` state is active.
Fixes#2310
Currently HttpConnector::set_local_address method accepts a single
argument. Server might not support IPv6 or IPv4. Therefore, the only
solution at the moment is to manually perform DNS resolution and pick
appropriate local address family. This is inefficient, as leads to
2 DNS lookups per request. This commit allows specifying both IPv4
and IPv6, so connector can decide which one to use based on DNS
resolution results.
Previously, calling `http1_writev(true)` would just keep the default behavior, which was to auto detect if writev was optimal. Now, the auto-detection is still default, but explicitly calling `http1_writev(true)` will skip the auto-detection, and always use writev queue strategy.
Closes#2282
Add basic, module-level example for the Http struct in the server::conn module that shows how to customize the configuration of the HTTP protocol handling and then serve requests received through a tokio TCP stream.
- update proto::h1::end_body to return Result<()>
- update Encoder::end to return Error(NotEof) only when there's Content-length left to be addressed
Closes#2263
I've moved Hyper from `log` to `tracing`. Existing `log`-based users shouldn't notice a difference, but `tracing` users will see higher performance when filtering data. This isn't the _end_ of the `tracing` integration that can happen in `Hyper` (e.g., Hyper can start using spans, typed fields, etc.), but _something_ is better than nothing. I'd rather address those points, including examples, in followups.
I've attached a screenshot of the `hello` example working, but the logged information is pulled from `tracing`, not `log`.
<img width="514" alt="Screen Shot 2020-05-16 at 1 23 19 PM" src="https://user-images.githubusercontent.com/2067774/82126298-d8103800-9779-11ea-8f0b-57c632c684d6.png">
Add `examples/service_struct_impl.rs`, which provides an up-to-date
example of implementing MakeService and Service on custom types.
The `Svc` struct has a Counter, which demonstrates how to instantiate
shared resources in the `MakeSvc` and pass the resource to the
services. Updates the `examples/README.md` and the doc in
`src/service/mod.rs`.
Closes#1691
If the runtime is dropped while a DNS request is being made, it can
lead to a panic. This patch checks if the task was cancelled, and
returns a generic IO error instead of panicking in that case.
The following code reproduces the problem on my macOS machine:
```
use hyper::Client;
use tokio::runtime;
type Result<T> = std::result::Result<T, Box<dyn std::error::Error + Send + Sync>>;
fn main() {
let rt = runtime::Builder::new()
.threaded_scheduler()
.core_threads(1)
.enable_all()
.build()
.unwrap();
// spawn a request and then drop the runtime immediately
rt.spawn(fetch_url());
}
async fn fetch_url() -> Result<()> {
let url: hyper::Uri = "http://example.com".parse()?;
let client = Client::builder().build_http::<hyper::Body>();
let res = client.get(url).await?;
println!("Response: {}", res.status());
Ok(())
}
```
A stream wrapped into a Body previously needed to implement `Sync` so
that the Body type implements this autotrait as well (which is needed
due to limitations in async/await). Since a stream only offers one
method that is called with an exclusive reference, this type is
statically proven to be Sync already. In theory it should be fine to add
an `unsafe impl Sync`, but this commit instead adds a SyncWrapper to
enlist the compiler’s help in proving that this is (and remains) correct.
This makes it easier to construct response bodies for client code.
